Merge branch 'previews' into develop

4 files changed, 148 insertions, 33 deletions

diff --git a/test/lib/test_bsie.py b/test/lib/test_bsie.py
index 38e6f59..ae23c4b 100644
--- a/test/lib/test_bsie.py
+++ b/test/lib/test_bsie.py
@@ -11,7 +11,7 @@ import unittest
 # bsie imports
 from bsie.extractor import ExtractorBuilder
 from bsie.extractor.base import SCHEMA_PREAMBLE
-from bsie.lib import PipelineBuilder
+from bsie.lib import PipelineBuilder, DefaultNamingPolicy
 from bsie.reader import ReaderBuilder
 from bsie.utils import bsfs, node, ns
 
@@ -40,13 +40,13 @@ class TestBSIE(unittest.TestCase):
                 )},
             ])
         # build pipeline
-        self.prefix = bsfs.Namespace('http://example.com/local/')
-        pbuild = PipelineBuilder(self.prefix, rbuild, ebuild)
+        self.naming_policy = DefaultNamingPolicy(host='http://example.com/local', user='')
+        pbuild = PipelineBuilder(rbuild, ebuild)
         self.pipeline = pbuild.build()
 
     def test_construction(self):
-        # pipeline only
-        lib = BSIE(self.pipeline)
+        # only pipeline and naming policy
+        lib = BSIE(self.pipeline, self.naming_policy)
         self.assertSetEqual(set(lib.principals), {
             ns.bse.filename,
             ns.bse.filesize,
@@ -70,7 +70,7 @@ class TestBSIE(unittest.TestCase):
             '''))
 
         # specify collect
-        lib = BSIE(self.pipeline, collect={
+        lib = BSIE(self.pipeline, self.naming_policy, collect={
             ns.bse.filesize,
             ns.bse.author,
             ns.bse.inexistent,
@@ -91,7 +91,7 @@ class TestBSIE(unittest.TestCase):
                 bsfs:unique "true"^^xsd:boolean .
             '''))
         # empty collect is disregarded
-        lib = BSIE(self.pipeline, collect={})
+        lib = BSIE(self.pipeline, self.naming_policy, collect={})
         self.assertSetEqual(set(lib.principals), {
             ns.bse.filename,
             ns.bse.filesize,
@@ -116,7 +116,7 @@ class TestBSIE(unittest.TestCase):
             '''))
 
         # specify discard
-        lib = BSIE(self.pipeline, discard={
+        lib = BSIE(self.pipeline, self.naming_policy, discard={
             ns.bse.filesize,
             ns.bse.filename,
             ns.bse.inexistent,
@@ -132,7 +132,7 @@ class TestBSIE(unittest.TestCase):
             '''))
 
         # specify collect and discard
-        lib = BSIE(self.pipeline,
+        lib = BSIE(self.pipeline, self.naming_policy,
             collect={ns.bse.filesize, ns.bse.author, ns.bse.foo, ns.bse.bar},
             discard={ns.bse.author, ns.bse.foo, ns.bse.foobar},
             )
@@ -150,14 +150,14 @@ class TestBSIE(unittest.TestCase):
 
     def test_from_file(self):
         # setup
-        lib = BSIE(self.pipeline)
+        lib = BSIE(self.pipeline, self.naming_policy)
         self.assertSetEqual(set(lib.principals), {
             ns.bse.filesize,
             ns.bse.filename,
             ns.bse.author,
             })
         content_hash = 'a948904f2f0f479b8f8197694b30184b0d2ed1c1cd2a1ec0fb85d299a192a447'
-        subject = node.Node(ns.bsfs.File, (self.prefix + 'file#')[content_hash])
+        subject = node.Node(ns.bsfs.File, uri=f'http://example.com/local/file#{content_hash}')
         testfile = os.path.join(os.path.dirname(__file__), 'testfile.t')
 
         # from_file extracts all available triples
diff --git a/test/lib/test_builder.py b/test/lib/test_builder.py
index 273d620..48e932b 100644
--- a/test/lib/test_builder.py
+++ b/test/lib/test_builder.py
@@ -21,7 +21,6 @@ from bsie.lib import PipelineBuilder
 
 class TestPipelineBuilder(unittest.TestCase):
     def test_build(self):
-        prefix = bsfs.URI('http://example.com/local/file#')
         c_schema = '''
             bse:author rdfs:subClassOf bsfs:Predicate ;
                 rdfs:domain bsfs:Entity ;
@@ -40,7 +39,7 @@ class TestPipelineBuilder(unittest.TestCase):
                 )},
             ])
         # build pipeline
-        builder = PipelineBuilder(prefix, rbuild, ebuild)
+        builder = PipelineBuilder(rbuild, ebuild)
         pipeline = builder.build()
         # delayed import
         import bsie.reader.path
@@ -61,7 +60,7 @@ class TestPipelineBuilder(unittest.TestCase):
             {'bsie.extractor.generic.path.Path': {}},
             ])
         with self.assertLogs(logging.getLogger('bsie.lib.builder'), logging.ERROR):
-            pipeline = PipelineBuilder(prefix, rbuild, ebuild_err).build()
+            pipeline = PipelineBuilder(rbuild, ebuild_err).build()
             self.assertDictEqual(pipeline._ext2rdr, {
                 bsie.extractor.generic.path.Path(): bsie.reader.path.Path()})
 
@@ -71,7 +70,7 @@ class TestPipelineBuilder(unittest.TestCase):
             {'bsie.extractor.generic.path.Path': {}},
             ])
         with self.assertLogs(logging.getLogger('bsie.lib.builder'), logging.ERROR):
-            pipeline = PipelineBuilder(prefix, rbuild, ebuild_err).build()
+            pipeline = PipelineBuilder(rbuild, ebuild_err).build()
             self.assertDictEqual(pipeline._ext2rdr, {
                 bsie.extractor.generic.path.Path(): bsie.reader.path.Path()})
 
@@ -81,7 +80,7 @@ class TestPipelineBuilder(unittest.TestCase):
             old_reader = bsie.extractor.generic.path.Path.CONTENT_READER
             bsie.extractor.generic.path.Path.CONTENT_READER = 'bsie.reader.foo.Foo'
             # build pipeline with invalid reader reference
-            pipeline = PipelineBuilder(prefix, rbuild, ebuild).build()
+            pipeline = PipelineBuilder(rbuild, ebuild).build()
             self.assertDictEqual(pipeline._ext2rdr, {
                 bsie.extractor.generic.stat.Stat(): bsie.reader.stat.Stat(),
                 bsie.extractor.generic.constant.Constant(c_schema, c_tuples): None,
@@ -92,7 +91,7 @@ class TestPipelineBuilder(unittest.TestCase):
         # fail to build reader
         rbuild_err = ReaderBuilder({'bsie.reader.stat.Stat': dict(foo=123)})
         with self.assertLogs(logging.getLogger('bsie.lib.builder'), logging.ERROR):
-            pipeline = PipelineBuilder(prefix, rbuild_err, ebuild).build()
+            pipeline = PipelineBuilder(rbuild_err, ebuild).build()
             self.assertDictEqual(pipeline._ext2rdr, {
                 bsie.extractor.generic.path.Path(): bsie.reader.path.Path(),
                 bsie.extractor.generic.constant.Constant(c_schema, c_tuples): None,
diff --git a/test/lib/test_naming_policy.py b/test/lib/test_naming_policy.py
new file mode 100644
index 0000000..4861c84
--- /dev/null
+++ b/test/lib/test_naming_policy.py
@@ -0,0 +1,120 @@
+"""
+
+Part of the bsie test suite.
+A copy of the license is provided with the project.
+Author: Matthias Baumgartner, 2022
+"""
+# standard imports
+import unittest
+
+# bsie imports
+from bsie.utils import ns, errors
+from bsie.utils.bsfs import URI
+from bsie.utils.node import Node
+
+# objects to test
+from bsie.lib.naming_policy import NamingPolicy, NamingPolicyIterator, DefaultNamingPolicy
+
+
+
+## code ##
+
+class TestDefaultNamingPolicy(unittest.TestCase):
+
+    def test_handle_node(self):
+        # setup
+        policy = DefaultNamingPolicy('http://example.com', 'me')
+        # handle_node doesn't modify existing uris
+        self.assertEqual(policy.handle_node(
+            Node(ns.bsfs.Entity, uri='http://example.com/you/foo#bar')).uri,
+            URI('http://example.com/you/foo#bar'))
+        # processes bsfs:File
+        self.assertEqual(policy.handle_node(
+            Node(ns.bsfs.File, ucid='abc123cba')).uri,
+            URI('http://example.com/me/file#abc123cba'))
+        # processes bsfs:Preview
+        self.assertEqual(policy.handle_node(
+            Node(ns.bsfs.Preview, ucid='abc123cba', size=123)).uri,
+            URI('http://example.com/me/preview#abc123cba_s123'))
+        # raises an exception on unknown types
+        self.assertRaises(errors.ProgrammingError, policy.handle_node,
+            Node(ns.bsfs.Entity, ucid='abc123cba', size=123))
+
+    def test_name_file(self):
+        # setup
+        policy = DefaultNamingPolicy('http://example.com', 'me')
+        # name_file uses ucid
+        self.assertEqual(policy.name_file(
+            Node(ns.bsfs.File, ucid='123abc321')).uri,
+            URI('http://example.com/me/file#123abc321'))
+        # name_file falls back to a random guid
+        self.assertTrue(policy.name_file(
+            Node(ns.bsfs.File)).uri.startswith('http://example.com/me/file#'))
+
+    def test_name_preview(self):
+        # setup
+        policy = DefaultNamingPolicy('http://example.com', 'me')
+        # name_preview uses ucid
+        self.assertEqual(policy.name_preview(
+            Node(ns.bsfs.Preview, ucid='123abc321')).uri,
+            URI('http://example.com/me/preview#123abc321'))
+        self.assertEqual(policy.name_preview(
+            Node(ns.bsfs.Preview, ucid='123abc321', size=400)).uri,
+            URI('http://example.com/me/preview#123abc321_s400'))
+        # name_preview uses source
+        self.assertEqual(policy.name_preview(
+            Node(ns.bsfs.Preview, source=Node(ns.bsfs.File, ucid='123file321'))).uri,
+            URI('http://example.com/me/preview#123file321'))
+        self.assertEqual(policy.name_preview(
+            Node(ns.bsfs.Preview, source=Node(ns.bsfs.File, ucid='123file321'), size=300)).uri,
+            URI('http://example.com/me/preview#123file321_s300'))
+        # name_preview falls back to a random guid
+        self.assertTrue(policy.name_preview(
+            Node(ns.bsfs.Preview)).uri.startswith('http://example.com/me/preview#'))
+        self.assertTrue(policy.name_preview(
+            Node(ns.bsfs.Preview, size=200)).uri.startswith('http://example.com/me/preview#'))
+        self.assertTrue(policy.name_preview(
+            Node(ns.bsfs.Preview, size=200)).uri.endswith('_s200'))
+
+
+class TestNamingPolicyIterator(unittest.TestCase):
+
+    def test_call(self): # NOTE: We test NamingPolicy.__call__ here
+        # setup
+        policy = DefaultNamingPolicy('http://example.com', 'me')
+        # call accepts list
+        triples = [('node', 'pred', 'value'), ('node', 'pred', 'value')]
+        it = policy(triples)
+        self.assertIsInstance(it, NamingPolicyIterator)
+        self.assertEqual(it._iterable, triples)
+        self.assertEqual(it._policy, policy)
+        # call accepts iterator
+        triples = iter([('node', 'pred', 'value'), ('node', 'pred', 'value')])
+        it = policy(triples)
+        self.assertIsInstance(it, NamingPolicyIterator)
+        self.assertEqual(it._iterable, triples)
+        self.assertEqual(it._policy, policy)
+
+    def test_iter(self):
+        # setup
+        policy = DefaultNamingPolicy('http://example.com', 'me')
+        triples = [
+            (Node(ns.bsfs.File, ucid='foo'), 'predA', 'hello'),
+            (Node(ns.bsfs.Preview, ucid='bar'), 'predB', 1234),
+            (Node(ns.bsfs.Preview, ucid='hello'), 'predC', Node(ns.bsfs.File, ucid='world'))
+            ]
+        # handles nodes, handles values, ignores predicate
+        self.assertListEqual(list(policy(triples)), [
+            (Node(ns.bsfs.File, uri='http://example.com/me/file#foo'), 'predA', 'hello'),
+            (Node(ns.bsfs.Preview, uri='http://example.com/me/preview#bar'), 'predB', 1234),
+            (Node(ns.bsfs.Preview, uri='http://example.com/me/preview#hello'), 'predC',
+                Node(ns.bsfs.File, uri='http://example.com/me/file#world')),
+            ])
+
+
+## main ##
+
+if __name__ == '__main__':
+    unittest.main()
+
+## EOF ##
diff --git a/test/lib/test_pipeline.py b/test/lib/test_pipeline.py
index 8fecc74..61fddd7 100644
--- a/test/lib/test_pipeline.py
+++ b/test/lib/test_pipeline.py
@@ -48,32 +48,28 @@ class TestPipeline(unittest.TestCase):
             bsie.extractor.generic.constant.Constant(csA, tupA): None,
             bsie.extractor.generic.constant.Constant(csB, tupB): None,
         }
-        self.prefix = bsfs.Namespace('http://example.com/local/')
 
     def test_essentials(self):
-        pipeline = Pipeline(self.prefix, self.ext2rdr)
+        pipeline = Pipeline(self.ext2rdr)
         self.assertEqual(str(pipeline), 'Pipeline')
         self.assertEqual(repr(pipeline), 'Pipeline(...)')
 
     def test_equality(self):
-        pipeline = Pipeline(self.prefix, self.ext2rdr)
+        pipeline = Pipeline(self.ext2rdr)
         # a pipeline is equivalent to itself
         self.assertEqual(pipeline, pipeline)
         self.assertEqual(hash(pipeline), hash(pipeline))
         # identical builds are equivalent
-        self.assertEqual(pipeline, Pipeline(self.prefix, self.ext2rdr))
-        self.assertEqual(hash(pipeline), hash(Pipeline(self.prefix, self.ext2rdr)))
+        self.assertEqual(pipeline, Pipeline(self.ext2rdr))
+        self.assertEqual(hash(pipeline), hash(Pipeline(self.ext2rdr)))
 
-        # equivalence respects prefix
-        self.assertNotEqual(pipeline, Pipeline(bsfs.URI('http://example.com/global/ent#'), self.ext2rdr))
-        self.assertNotEqual(hash(pipeline), hash(Pipeline(bsfs.URI('http://example.com/global/ent#'), self.ext2rdr)))
         # equivalence respects extractors/readers
         ext2rdr = {ext: rdr for idx, (ext, rdr) in enumerate(self.ext2rdr.items()) if idx % 2 == 0}
-        self.assertNotEqual(pipeline, Pipeline(self.prefix, ext2rdr))
-        self.assertNotEqual(hash(pipeline), hash(Pipeline(self.prefix, ext2rdr)))
+        self.assertNotEqual(pipeline, Pipeline(ext2rdr))
+        self.assertNotEqual(hash(pipeline), hash(Pipeline(ext2rdr)))
 
         # equivalence respects schema
-        p2 = Pipeline(self.prefix, self.ext2rdr)
+        p2 = Pipeline(self.ext2rdr)
         p2._schema = bsfs.schema.Schema()
         self.assertNotEqual(pipeline, p2)
         self.assertNotEqual(hash(pipeline), hash(p2))
@@ -90,10 +86,10 @@ class TestPipeline(unittest.TestCase):
 
     def test_call(self):
         # build pipeline
-        pipeline = Pipeline(self.prefix, self.ext2rdr)
+        pipeline = Pipeline(self.ext2rdr)
         # build objects for tests
         content_hash = 'a948904f2f0f479b8f8197694b30184b0d2ed1c1cd2a1ec0fb85d299a192a447'
-        subject = node.Node(ns.bsfs.File, (self.prefix + 'file#')[content_hash])
+        subject = node.Node(ns.bsfs.File, ucid=content_hash)
         testfile = os.path.join(os.path.dirname(__file__), 'testfile.t')
         p_filename = pipeline.schema.predicate(ns.bse.filename)
         p_filesize = pipeline.schema.predicate(ns.bse.filesize)
@@ -138,7 +134,7 @@ class TestPipeline(unittest.TestCase):
             def __call__(self, path):
                 raise errors.ReaderError('reader error')
 
-        pipeline = Pipeline(self.prefix, {bsie.extractor.generic.path.Path(): FaultyReader()})
+        pipeline = Pipeline({bsie.extractor.generic.path.Path(): FaultyReader()})
         with self.assertLogs(logging.getLogger('bsie.lib.pipeline'), logging.ERROR):
             testfile = os.path.join(os.path.dirname(__file__), 'testfile.t')
             p_filename = pipeline.schema.predicate(ns.bse.filename)
@@ -149,7 +145,7 @@ class TestPipeline(unittest.TestCase):
             def extract(self, subject, content, predicates):
                 raise errors.ExtractorError('extractor error')
 
-        pipeline = Pipeline(self.prefix, {FaultyExtractor(): bsie.reader.path.Path()})
+        pipeline = Pipeline({FaultyExtractor(): bsie.reader.path.Path()})
         with self.assertLogs(logging.getLogger('bsie.lib.pipeline'), logging.ERROR):
             testfile = os.path.join(os.path.dirname(__file__), 'testfile.t')
             p_filename = pipeline.schema.predicate(ns.bse.filename)
@@ -157,7 +153,7 @@ class TestPipeline(unittest.TestCase):
 
     def test_predicates(self):
         # build pipeline
-        pipeline = Pipeline(self.prefix, self.ext2rdr)
+        pipeline = Pipeline(self.ext2rdr)
         #
         self.assertSetEqual(set(pipeline.principals), {
             pipeline.schema.predicate(ns.bse.filename),